Tumor and Salivary Matrix Metalloproteinase Levels Are Strong Diagnostic Markers of Oral Squamous Cell Carcinoma

Published OnlineFirst September 29, 2011; DOI: 10.1158/1055-9965.EPI-11-0503

Cancer Epidemiology, Research Article Biomarkers & Prevention

Marni Stott-Miller1,2, John R. Houck1, Pawadee Lohavanichbutr1, Eduardo Mendez 1,3,5, Melissa P. Upton4, Neal D. Futran3, Stephen M. Schwartz1,2, and Chu Chen1,2,3

Abstract Background: The matrix metalloproteinases (MMP) cause degradation of the extracellular matrix and basement membranes, and thus may play a key role in cancer development. Methods: In our search for biomarkers for oral squamous cell carcinomas (OSCC), we compared primary OSCC, oral dysplasia and control subjects with respect to: (i) expression of MMP1, MMP3, MMP10, and MMP12 in oral epithelial tissue using Affymetrix U133 2.0 Plus GeneChip arrays, followed by quantitative reverse transcription-PCR (qRT-PCR) for MMP1, and (ii) determination of MMP1 and MMP3 concentrations in saliva. Results: MMP1 expression in primary OSCC (n ¼ 119) was >200-fold higher (P ¼ 7.16 10 40) compared with expression levels in nonneoplastic oral epithelium from controls (n ¼ 35). qRT-PCR results on 30 cases and 22 controls conﬁrmed this substantial differential expression. The exceptional discriminatory power to separate OSCC from controls was validated in two independent testing sets (AUC% ¼ 100; 95% CI: 100–100 and AUC% ¼ 98.4; 95% CI: 95.6–100). Salivary concentrations of MMP1 and MMP3 in OSCC patients (33 stage I/II, 26 stage III/IV) were 6.2 times (95% CI: 3.32–11.73) and 14.8 times (95% CI: 6.75–32.56) higher, respectively, than in controls, and displayed an increasing trend with higher stage disease. Conclusion: Tumor and salivary MMPs are robust diagnostic biomarkers of OSCC. Impact: The capacity of MMP gene expression to identify OSCC provides support for further inves- tigation into MMPs as potential markers for OSCC development. Detection of MMP proteins in saliva in particular may provide a promising means to detect and monitor OSCC noninvasively. Cancer Epidemiol Biomarkers Prev; 20(12); 2628–36. 2011 AACR.

Introduction 5-year survival is estimated at 63% for Whites and 43% for Blacks in the Unites States (2). Prognosis is heavily based Oral squamous cell carcinoma (OSCC) of the oral cavity on the AJCC staging system, which has limited utility for and oropharynx is one of the most common cancers in the predicting survival because patients with tumors of the world, with an estimated 400,000 new cases and 200,000 same AJCC stage often have heterogeneous responses to deaths in 2008 worldwide (1). In 2010, an estimated 36,540 treatment (4). new cases and 7,880 deaths from OSCC occurred in the To improve the mortality and morbidity burden asso- United States (2). Despite advances in chemotherapy, ciated with OSCC, there is an urgent need to identify radiation treatments, screening tools such as the VEL- sensitive and speciﬁc biomarkers for early detection and scope (LED Dental, Inc.; ref. 3), and improvements in prognosis. For utmost clinical utility, these biomarkers surgical techniques, survival remains extremely poor. The should be measurable in specimens that are collected with minimal discomfort to the patient (5).

Authors' Afﬁliations: 1Program in Epidemiology, Fred Hutchinson Cancer The matrix metalloproteinases (MMP) have frequent- Research Center; Departments of 2Epidemiology, 3Otolaryngology-Head ly been studied as potential cancer biomarkers, includ- and Neck Surgery, and 4Pathology, University of Washington; and 5Surgery ing in OSCC, and have been associated with tumor and Perioperative Care Service, VA Puget Sound Health Care System, Seattle, Washington invasion and metastases (6–9). Insertion/deletion (ins/del) polymorphisms in the promoter region of Note: Supplementary data for this article are available at Cancer Epide- miology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/). MMP1 ( 1,607, 1G/2G, rs1799750; ref. 10) and MMP3 (1,171, 5A/6A, rs3025039; ref. 11) inﬂuence the level of Corresponding Author: Chu Chen, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, M5-C800 P.O. Box 19024, Seattle, MMP expression. However, epidemiologic studies WA 98109. Phone: 206-667-6644; Fax: 206-667-2537; E-mail: investigating these polymorphisms and oral cancer risk [email protected] have not been entirely consistent (12–14). Overexpres- doi: 10.1158/1055-9965.EPI-11-0503 sion of MMP transcripts in OSCC, have been observed 2011 American Association for Cancer Research. repeatedly (7, 15–18). The protein concentrations of

2628 Cancer Epidemiol Biomarkers Prev; 20(12) December 2011

Tumor and Salivary Matrix Metalloproteinases in Oral Cancer

MMP1, MMP2, MMP3, and MMP9 have been observed patients and controls was obtained at time of biopsy or to be higher in OSCC tumor tissue compared with surgery. Oral epithelial tissue from controls was collect- control tissue (19). However, many of these studies are ed from the uvula or anterior tonsillar pillar, avoiding limited by having very small numbers of patients contamination with surrounding lymphoid tissue. The (15, 16) or by use of control tissue from OSCC patients removed tissue was immersed in RNALater (Applied instead of OSCC-free controls (7, 17–19). MMP1 tran- Biosystems, Inc.) for at least 12 hours at 4C before storage script levels in saliva have been shown to distinguish at 80C until use. OSCCcasesfromcontrols(AUC¼ 98.9%; ref. 20). In Saliva samples were collected preoperatively during addition, overexpression of MMP1 and MMP9 have the clinic visit. Patients were asked not to eat or drink been associated with progression of dysplasia to cancer for at least 1 hour before collection and to spit into a (21). centrifuge tube. The saliva was stored at 4C for up to This study sought to focus on the MMPs as potential 2 hours, then centrifuged for 10 minutes at 1,300 g and diagnostic and prognostic biomarkers. Our primary goal aliquoted into cryovials for storage at 80C. was to determine whether salivary concentrations of the most highly differentially expressed MMPs might be Patient follow-up useful as a diagnostic aid. Although others have Patient follow-up was done from July, 2004 to January, reported elevated MMP transcript levels in saliva (20), 2011. Subjects were followed through periodic tele- protein biomarkers in saliva should provide more accu- phone contacts at as close to 6-month intervals as pos- rate representations of molecular function. There are sible following surgery, review of medical records and numerous posttranscriptional processes, including tran- linkage to data sources. Vital status was checked against script de/stabilization, translation, protein modification, the Social Security Death Index (SSDI) and Fred Hutch- and degradation. As such, mRNA levels may not cor- inson Cancer Research Center’s Cancer Surveillance relate with protein activity (22, 23). In addition, saliva System (CSS), which is part of the Surveillance, Epide- is an ideal diagnostic tool for biomarker assessment due miology, and End Results (SEER) program of the Nation- to its accessibility and ease of collection (5, 24). An al Cancer Institute, and is updated with the Washington additional goal was to evaluate the use of the MMPs as State Death Certificate database and National Death a prognostic aid by investigating whether MMP expres- Index. A death was classified as due to OSCC or not sion was associated with an increased risk of death. due to OSCC based on review of medical records and death certificates by head and neck surgeons involved in Methods thestudy.Iftherewasnoindicationofdeath,wecen- soredthatsubjectatthelastknowndateoffollow-up. Study population Only 2 subjects were lost to follow-up. As described by Chen and colleagues (25), cases were previously untreated primary OSCC patients scheduled Laboratory studies for biopsy or surgery at the University of Washington As described previously (25), total RNA was extracted Medical Center, Harborview Medical Center, or Veterans from tumor and nonneoplastic oral epithelium using a Affairs (VA) Puget Sound Health Care System. Patients TRIzol method (Invitrogen), purified with an RNeasy diagnosed with oral dysplastic lesions were also enrolled mini kit (Qiagen), processed using a GeneChip Expression through these medical centers during the same time 30-Amplification Reagents Kit (Affymetrix), and exam- period. Controls were patients who received oral surgery, ined with Affymetrix GeneChip Human Genome U133 such as uvulopalatopharyngoplasty and tonsillectomy, Plus 2.0 Arrays. for a nonmalignant or nonpremalignant condition, at Tumor and nonneoplastic oral epithelium samples the same institutions and during the same time period. were screened for the presence of HPV DNA using a Subjects were enrolled between December 2003 and April nested PCR protocol (26). All samples that showed a 2007. positive result were tested for HPV types using the LIN- Study participants were interviewed using a struc- EAR ARRAY HPV Genotyping Test (Roche), under a tured questionnaire eliciting demographic characteris- research-use only agreement. "High-risk" HPV was tics, medical and lifestyle history, including tobacco and defined by 13 types, including HPV-16 and HPV-18; alcohol use. Data on tumor stage and other character- "low-risk" HPV corresponded to 24 types. To verify HPV istics were abstracted from medical records. Participants type calls, a subset of the samples were amplified and gave informed consent and study procedures were sequenced using HPV-16–specific primers, and compared approved by the Institutional Review Boards of the Fred against a known HPV-16 sequence (GenBank 333031). Hutchinson Cancer Research Center, University of We verified MMP1 gene expression results by quanti- Washington, and VA Puget Sound Health Care System. tative reverse transcription-PCR (qRT-PCR) on a subset of 30 OSCC cases and 22 controls using a QuantiTect SYBR Tissue and saliva collection Green RT-PCR kit (Qiagen) and bioinformatically validat- Tumor tissue from cases was obtained at time of resec- ed QuantiTect primers (Qiagen) on an ABI 7900HT tion prior to other treatment. Tissue from dysplasia Sequence Detection System (Applied Biosystems, Inc.).

www.aacrjournals.org Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 2629

Stott-Miller et al.

Standard curves were generated using Universal Human ANCOVA (adjusting for age and sex) and a Student t test Reference RNA (Stratagene) for all genes, with the linear using Partek Genomics Suite. We set the false discovery correlation coefficient (r2) 0.99 for all runs. The mean rate at 1%. We then created a list for further analyses that threshold cycle (Ct) values were calculated from triplicate contained only those probe sets with (i) t score greater > Ct values. Samples that had Ct values with a SD 0.3 in the than 6, and (ii) a 1.5-fold or greater difference in gene triplicate run were repeated. Mean Ct values were stan- expression between cases and controls (on the log2 scale). b dardized to the mean Ct value of the referencegene -actin. We identified the top MMPs by sorting this list according The cycling conditions consisted of 30 minutes at 50C, to case–control fold-difference. 15 minutes at 95C, and 40 cycles each of 15 seconds at We compared expression levels of these top differen- 94C, 30 seconds at 55C, and 30 seconds at 72C. tially expressed MMPs in primary OSCC (n ¼ 119) with Saliva samples from 100 subjects (60 primary OSCC, expression levels in oral epithelium from control subjects 15 dysplasia, and 25 controls) were tested for MMP1 and (n ¼ 35) by fitting linear regression models, adjusted for MMP3 concentration. These MMPs were selected because age and sex, and estimating 95% CI for all association they exhibited the largest fold-difference out of MMPs estimates. We repeated these analyses with additional analyzed in gene expression analyses. Each sample adjustment for pack years of smoking (continuous) and in duplicate was analyzed blinded to patient status by alcoholic drinks per day in the year prior to the date of Aushon Biosciences Inc. using a SearchLight multiplex diagnosis for OSCC cases or recruitment for controls sandwich-ELISA proteome array (Aushon BioSystems), (categorical). We calculated Pearson correlation coeffi- with images analyzed using the SearchLight Array cients for the top 4 differentially expressed MMPs. Analyst software. Protein concentrations from 2 subjects In secondary analyses using data from both the train- could not be measured because of excessive viscosity. ing and testing sets, we conducted linear regression analyses of MMP1 case–control differences, adjusting MMP expression analyses for primary OSCC for age and sex, after separating by (i) site: oral cavity compared with control epithelium (115 cases, 45 controls) versus oropharynx (52 cases, Two rounds of quality control checks were done on the 45 controls), and (ii) HPV status: high-risk HPV positive microarray gene expression data. First, recommendations (56 cases, 4 controls) versus HPV negative or low-risk by Affymetrix were followed to determine if any Gene- HPV (111 cases, 40 controls). Chips needed to be excluded (27). Second, the "affyQCRe- port" and "affyPLM" software in the Bioconductor pack- Validation of MMP1 results using independent age within R statistical programming language were used testing sets to identify any poor-quality chips (28). In total, 167 cases We extracted gene expression values from files of our and 45 controls passed both quality control procedures. independent testing set (48 OSCC cases and 10 controls) The data were divided into a training set (119 cases and and a publicly available data set from Gene Expression 35 controls) and a testing set (48 cases and 10 controls). Omnibus (GEO; GSE6791; 42 OSCC cases and 14 controls; We used the training set in analyses that identified the ref. 30), and normalized the 2 data sets using the RMA top differentially expressed genes, validated results in an algorithm. We then used the receiver operator character- independent testing set to evaluate discriminatory capac- istic (ROC) curve to calculate, for each data set, the area ity of identified genes, and then merged the training and under the ROC curve (AUC) to evaluate the discrimina- testing sets to provide greater numbers of subjects for tory performance of the statistical prediction models. secondary analyses on stage, survival, and stratified by site. We extracted gene expression values for 54,000 Statistical Analyses probe sets in the training set from CEL files and normalized the data using the RMA algorithm in Partek Salivary MMP1 and MMP3 concentration Genomics Suite (29). We log-transformed the MMP1 and MMP3 concen- Although overexpression of the MMPs in OSCC had trations to correct for skewed data. We calculated the previously been reported by others (6–9, 15–19), our prior ratio of geometric means of their concentrations in cases gene expression analyses did not identify the MMPs (n ¼ 59) and control subjects (n ¼ 25) by fitting adjusted among the list of top differentially expressed genes regression models, adjusted for age and sex. We then because we had excluded probe sets based on magnitude repeated these analyses for MMP1 after stratifying by (probe sets were excluded if the expression value for that site: oral cavity (45 cases, 25 controls) versus orophar- < probe set in any of the samples was 3 on log2 scale; ynx (15 cases, 25 controls), and HPV status: high-risk ref. 25). This study did not eliminate such probe sets. HPV positive (15 cases, 1 control) versus high-risk HPV Using the training set, we eliminated probe sets if expres- negative (43 cases, 24 controls). sion showed little variation (defined as interquartile range In secondary analyses, we calculated the geometric of expression levels less than 0.3 on the log2 scale), result- mean for control, dysplasia, stage I/II OSCC, and stage ing in 35,000 probe sets for further analyses. III/IV OSCC. To evaluate sensitivity and specificity, we To compare expression levels of tumor tissue to tissue conducted ROC analyses and calculated AUC for salivary from control subjects in the training set, we conducted MMP1 and MMP3. To determine whether tissue mRNA

2630 Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 Cancer Epidemiology, Biomarkers & Prevention

Tumor and Salivary Matrix Metalloproteinases in Oral Cancer

levels of MMP1 and MMP3 correlated with corresponding The difference in mean log2 expression levels between salivary concentrations, we calculated the Pearson corre- primary OSCC and epithelium from cancer-free controls lation coefficient. in the training set, adjusting for age and sex, was 7.74 In all analyses for which software is not specified, we (95% CI: 7.05–8.44; Table 2) for MMP1, 5.30 (95% CI: used STATA statistical software (version 10.0, Stata Corp.) 4.57–6.03) for MMP3, 5.19 (95% CI: 4.52–5.84) for MMP10, and 5.26 (95% CI: 4.47–6.06) for MMP12. Results were Associations of MMP1 and MMP3 gene expression similar for analyses that additionally adjusted for pack with stage and survival years of smoking and alcoholic drinks per day (results not We merged the training and testing datasets to compare shown). The Pearson correlation coefficients between the expression levels of MMP1 and MMP3 (the top 2 differ- 4 MMPs ranged from 0.72 to 0.92 (P < 0.001, Table 3). entially expressed genes based on the initial microarray Using the combined training and testing sets, after analyses) across different disease categories. We were stratifying by site and adjusting for age and sex, the dif- unable to categorize 2 subjects because of missing data. ference in means for MMP1 (compared with control oral After categorization, we calculated mean log2 MMP1 and epithelium) was somewhat higher for oral cavity cancer MMP3 expression levels for 45 controls, 17 dysplasia, 54 (difference in means, 8.32; 95% CI: 7.75–8.90), than for stage I/II primary OSCC, and 111 stage III/IV primary oropharyngeal cancer (6.04; 95% CI: 5.09–6.99). Similarly, OSCC. the difference in means was higher for OSCC among We conducted survival analyses for primary OSCC high-risk HPV-negative subjects (difference in means, cases using data from our training and testing sets. The 8.30; 95% CI: 7.70–8.90), than among high-risk HPV-pos- primary outcome variable was survival time, measured as itive subjects (6.00; 95% CI: 3.99–8.02). For oropharyngeal days after diagnosis until a death occurred, or until sub- cancers only, the difference in means was 6.38 (95% CI: jects were lost to follow-up, or January 25, 2011. We started 4.54–8.22) among high-risk HPV-negative subjects and 6.10 the accumulation of follow-up time and deaths at 4 months (95% CI: 3.90–8.31) among high-risk HPV positive subjects. after diagnosis date to exclude any deaths due to treat- Similar patterns were observed for the other MMPs. ment-related complications. There were 159 subjects with Results of ROC analyses on our testing set (48 cases, at least 4 months of follow-up time. To analyze risk of 10 controls) indicated that MMP1 had exceptional dis- death based on gene expression values of MMP1 and criminatory power to separate OSCC cases from controls MMP3, we calculated OSCC-specific and all cause HR by (AUC ¼ 100%; 95% CI: 100–100). The predictive accuracy Cox proportional hazards regression, adjusting for age, of MMP1 was confirmed in the GEO dataset, GSE6791 sex, and stage and HPV status. We then conducted mutual (AUC ¼ 98.4%; 95% CI: 95.6–100). adjustment for MMP1 and MMP3. In the combined Affymetrix array data from the training set and our testing set, the mean of the log2 Results MMP1 expression levels for 45 control oral epithelium was 3.49 (95% CI: 3.21–3.78), 17 dysplasia samples was Cases tended to be older than control patients, and 6.69 (95% CI: 5.21–8.17), 54 stage I/II primary OSCC was were more likely to be male, White and current smokers 11.43 (95% CI: 10.94–11.93) and 111 stage III/IV OSCC (Table 1). In the training set, 74% of cases had oral cavity was 10.92 (95% CI: 10.46–11.38, Fig. 1). The pattern for tumors and 26% had oropharyngeal tumors. In our inter- MMP3 expression was similar (Fig. 1). nal independent testing set, oral cavity tumors accounted Our qRT-PCR results confirmed the substantial differ- for 60% and oropharyngeal tumors for 40% of the cases. ential expression of MMP1, with a difference in mean Ct value for controls versus cases of 10.43 (95% CI: 9.31–11.55, Gene expression levels of MMPs P < 0.001). The Pearson correlation coefficient for the Results of ANCOVA analyses on the training set microarray gene expression values and the qRT-PCR resulted in 16,228 significant probe sets out of a total results was 0.99 (P < 0.001). of 33,057. After preprocessing and filtering, we identified 173 probe sets that were substantially differentially Salivary MMP1 and MMP3 concentration expressed between cases and controls (Supplementary After log-transformation, salivary concentrations Table S1). After sorting by fold-difference (cases vs. of MMP1 for primary OSCC patients (n ¼ 59) were controls), the top overexpressed genes in this list were 6.2 times higher than in control patients (n ¼ 25), MMP1, MMP3, MMP10,andMMP12. MMP1 had an adjusting for age and sex (95% CI: 3.32–11.73; P < exceptionally high fold-difference (>200-fold; 3.16-fold 0.001). For MMP3, adjusted concentrations were 14.8 on log2 scale) and very low adjusted ANCOVA P value times higher in cases than in control patients (95% CI: (P ¼ 7.16 10 40). We observed significant differential 6.75–32.56; P < 0.001). Gene expression values in tissue expression for multiple additional MMPs, including correlated well with salivary concentrations for MMP1 MMP2 (3.5-fold-difference; P ¼ 2.01 10 20) and MMP9 (Pearson correlation coefficient, 0.63; P < 0.001) and (8.0-fold-difference; P ¼ 2.29 10 13), but elected to MMP3 (0.67; P < 0.001). conduct further analyses only on the 4 MMPs that were In stratified analyses of MMP1 salivary concentra- at the top of our list. tions, results were stronger for patients with oral cavity

www.aacrjournals.org Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 2631

Stott-Miller et al.

Table 1. Selected characteristics of OSCC patients and controls, University of Washington Afﬁliated Hospitals, 2003 to 2007

Training set Testing set

OSCC case Control OSCC case Control (n ¼ 119) (n ¼ 35) (n ¼ 48) (n ¼ 10) n % n % n % n %

Age (y) 19–39 5 (4.2) 14 (40.0) 2 (4.2) 3 (30.0) 40–49 18 (15.1) 8 (22.9) 8 (16.7) 6 (60.0) 50–59 40 (33.6) 4 (11.4) 17 (35.4) 1 (10.0) 60–88 56 (47.1) 9 (25.7) 21 (43.8) 0 (0.0) Sex Male 84 (70.6) 25 (71.4) 36 (75.0) 7 (70.0) Female 35 (29.4) 10 (28.6) 12 (25.0) 3 (30.0) Race White 106 (93.0) 24 (68.6) 40 (85.1) 6 (66.7) Non-White 8 (7.0) 11 (31.4) 7 (14.9) 3 (33.3) Unknown 5 (0) 1 (1.0) 0 (0) 0 (0) Cigarette smoking statusa Never/former 59 (49.6) 25 (71.4) 27 (56.3) 8 (80.0) Current 60 (50.4) 10 (28.6) 21 (43.7) 2 (20.0) Alcoholic drinking statusa Never/former 36 (30.8) 9 (25.7) 19 (40.4) 3 (30.0) Current 81 (69.2) 26 (74.3) 28 (59.6) 7 (70.0) Unknown 2 (0) 1 (0) 0 (0) 0 (0) AJCC stage I/II 42 (35.3) 13 (27.1) III/IV 77 (64.7) 35 (72.9) Tumor site Oral cavity 87 (73.1) 28 (58.3) Oropharynx 32 (26.9) 20 (41.7) High risk HPV status, by tumor siteb Oral cavity HPVþ 14 (16.1) 5 (17.9) HPV 73 (83.9) 23 (82.1) Oropharynx HPVþ 25 (78.1) 12 (60.0) HPV 7 (21.9) 8 (40.0)

aAs of the date of diagnosis (OSCC cases) or recruitment (controls). bHigh risk samples contained HPV-16, HPV-35, or HPV-45. Percentages estimated within site.

cancers compared with controls (ratio of geometric (Fig. 2). However, differences between categories did not means, 8.86; 95% CI: 1.54–16.82) versus patients with reach statistical signiﬁcance. oropharyngeal cancers compared with controls (2.51; In ROC analyses, the AUC was 84.5% (95% CI: 76.00– 95% CI: 0.90–6.97). Similarly, results were stronger 92.95%) for MMP1 and 87.66% (95% CI: 80.16–95.17%) for among high-risk HPV-negative subjects (ratio of geo- MMP3. A model that included both MMP1 and MMP3 did metric means, 6.79; 95% CI: 3.40–13.56) versus high-risk not improve upon the AUC obtained for either MMP1 or HPV-positive subjects (5.86; 95% CI: 2.07–16.55). The MMP3 alone. pattern for MMP3 was similar. In analyses across disease categories (controls, dyspla- Associations of MMP1 and MMP3 transcript levels sia, stage I/II OSCC, and stage III/IV OSCC), the geomet- with survival ric means of salivary MMP1 and MMP3 concentrations The median follow-up time for primary OSCC cases ¼ displayed an increasing trend with higher stage of disease (n 159) was 51 months. For each unit increase in log2

2632 Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 Cancer Epidemiology, Biomarkers & Prevention

Tumor and Salivary Matrix Metalloproteinases in Oral Cancer

Table 2. Difference in mean MMP expression levels between OSCC cases and controls, University of Washington Afﬁliated Hospitals, 2003 to 2007

mRNAb (119 cases, 35 controls) Protein (pg/mL;c 59 cases, 25 controls)

Gene/protein Difference in 95% CI Ratio of geometric 95% CI a a log2 means means MMP1 7.74 7.05–8.44 6.24 3.32–11.73 MMP3 5.30 4.57–6.03 14.83 6.75–32.56 MMP10d 5.19 4.52–5.84 —d — MMP12d 5.26 4.47–6.06 —d —

aAdjusted for age and sex. bMeasured in OSCC tissue of cases and oral epithelium of controls. cMeasured in saliva of OSCC cases and controls. dProtein concentrations not measured in saliva.

expression value, the HR for MMP1,adjustedforage, stitial collagen, such as collagen types I, II, III, and IV sex, stage, and HPV status was 1.24 (95% CI: 1.02–1.52) (34, 35). Among the MMPs, MMP1 is the most ubiqui- for OSCC-specific death and 1.20 (95% CI: 1.03–1.40) tously expressed interstitial collagenase and plays a for death from all causes. The adjusted HR was 1.09 key role in initial cleavage of the extracellular matrix. (0.86–1.38) for death from known non-OSCC causes. The adjusted HR for MMP3 expression for each unit increase in log2 value was 1.26 (95% CI: 1.09–1.46) for MMP1 OSCC-specific death and 1.25 (95% CI: 1.10–1.41) for death from all causes. We obtained similar results in analyses with adjustment for age and sex only. HRs for OSCC-specific death were lowered for MMP1 expression levels that were additionally adjusted for 10 15 MMP3 expression (HR, 0.96; 95% CI: 0.62–1.48), and were increased for MMP3 expression values that were additionally adjusted for MMP1 expression (HR, 1.30; 95% CI: MMP1 expression levels MMP1 expression 0.89–1.92). These patterns were similar for all-cause 2 mortality. log 0 Discussion Control Dysplasia Stage I/II OSCC Stage III/IV OSCC MMP3 45

The MMPs are believed to play a key role in tumor 1 metastasis, cell migration, cancer cell growth, and angio- genesis (31–33). They promote cell invasion by cleaving 2 components of the extracellular matrix, with MMP col- lagenases having the unique ability to cleave native inter- 1086 MMP3 expression levels MMP3 expression Table 3. Pearson correlation coefﬁcients 2 between MMP1, MMP3, MMP10, and MMP12a log 41

Control Dysplasia Stage I/II OSCC Stage III/IV OSCC MMP1 MMP3 MMP10 MMP12

MMP1 1.00 Figure 1. log2 MMP1 and MMP3 expression levels across disease MMP3 0.92 1.00 categories, University of Washington Afﬁliated Hospitals, 2003 to 2007. MMP10 0.89 0.85 1.00 MMP1 and MMP3 mRNA expression levels were measured from MMP12 0.84 0.76 0.72 1.00 Affymetrix microarray data, and are expressed as expression levels in the log2 scale. Plots are shown for expression levels in oral epithelium for: 45 aAll correlations P < 0.001. controls, 17 dysplasia patients, 54 stage I/II invasive OSCC patients, and 111 stage III/IV invasive OSCC patients.

www.aacrjournals.org Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 2633

Stott-Miller et al.

The gene expression levels of the MMPs were highly MMP1 correlated. In particular, we observed a correlation of 0.92 (P < 0.001) for MMP1 and MMP3. Therefore, although each of these genes is a strong biomarker for OSCC, the

10 12 combination of the 2 does not add to their predictive ability. MMP1 and MMP3 are both on chromosome 11 in the region 11q22.3 (36). They are often observed to be coordinately expressed (37, 38), and are activated by similar factors, such as interleukin-1 (39). 64 Protein concentrations of both MMP1 and MMP3 were observed to be highly elevated in the saliva of OSCC patients compared with saliva from cancer-free controls. log-transformed protein concentration (pg/mL) protein concentration log-transformed Control Dysplasia Stage I/II Stage III/IV Although results did not reach statistical signiﬁcance, we observed a trend toward higher concentrations for both MMP3

0 MMP1 and MMP3 with increasing disease severity. Sev- eral studies have evaluated salivary proteins as potential diagnostic tools in OSCC (24, 40–43). Hu and colleagues 86 identiﬁed a panel of protein biomarkers in the human salivary proteome and observed that a combination of 5 proteins yielded an AUC of 93% for OSCC detection (40). Shpitzer and colleagues observed increased salivary protein concentrations of 75% for MMP2 and 35% for 48 MMP9 in cases versus controls (42, 43). The MMPs inves- tigated in our study were not among the proteins reported

21 in these previous studies. log-transformed protein concentration (pg/mL) protein concentration log-transformed Control Dysplasia Stage I/II Stage III/IV Oral cavity and oropharyngeal squamous cell carcinoma are particularly suitable for salivary biomarker mea-

Figure 2. MMP1 and MMP3 protein concentration levels in saliva across surements due to the presence of potentially abnormal disease categories, University of Washington Affiliated Hospitals, 2003 cells/markers sloughed off directly into saliva. Identifi- to 2007. MMP1 and MMP3 protein concentrations were measured in cation of a sensitive and specific marker of OSCC by saliva, and are expressed as log-transformed protein concentrations in noninvasive means such as salivary analysis has great pg/mL. Plots are shown for protein concentrations in saliva for: 25 potential to assist diagnosis and prognosis (5). People at controls, 14 dysplasia patients, 33 stage I/II invasive OSCC patients, and 26 stage III/IV invasive OSCC patients. high risk of developing OSCC, such as heavy smokers, could potentially be monitored for the first indications of OSCC development. Salivary marker analysis could be done in between biopsies to assist in the monitoring of Results of this study highlight the importance of the disease status of dysplasia patients (24). Salivary moni- MMPs in the development of OSCC. The prominence of toring has the advantage of being less invasive, and these genes is concordant with similar microarray experi- provides a mechanism to possibly detect lesions in loca- ments; for example, Ziober and colleagues (7) reported tions that are difficult to be visualized by general exam- MMPs on their list of 25 genes overexpressed in OSCC. In ination. In addition, the best strategy for management of particular, MMP1 emerged as an exceptionally strong patients presenting with mild to moderate dysplasia or marker in our study. This gene was expressed at a level with other atypical lesions is currently unclear. Markers more than 200-fold greater in primary OSCC compared associated with invasive disease (in saliva or the biopsy with oral epithelium from controls. Importantly, MMP1 itself) could assist clinicians in determining which perfectly distinguished OSCC from control tissue in ROC patients should undergo further screening or more inva- analyses in our independent testing set (AUC ¼ 100%), sive treatment. and almost perfectly discriminated OSCC from control We obtained different results after stratification by site, tissue in an external independent testing set (AUC ¼ with greater case–control differences for MMP gene 98.4%). The substantial differential expression of MMP1 expression levels and salivary protein concentrations in was confirmed by qRT-PCR. In addition, we analyzed oral cavity cancer versus oropharyngeal cancer. Similarly, expression of MMP1 and MMP3 across different stages in case–control differences for MMP gene expression levels the natural history of OSCC and observed increasing and salivary concentrations were greater among high-risk expression corresponding to progression from normal HPV-negative subjects versus high-risk HPV positive tissue to dysplasia to OSCC (although not from stage I/ subjects, although the very small number of high-risk II to stage III/IV OSCC). The MMPs may thus be ideal HPV-positive controls limit meaningful interpretation of molecular markers for monitoring progression from dys- these results. This pattern is congruent with results plasia to invasive OSCC. obtained after stratification by site, as cancers occurring

2634 Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 Cancer Epidemiology, Biomarkers & Prevention

Tumor and Salivary Matrix Metalloproteinases in Oral Cancer

in the oropharynx, as opposed to the oral cavity, are most the underlying network of biological processes that are strongly associated with HPV infection (44, 45). Higher involved. Unlike the situation for MMPs, a commercially MMP expression in HPV-negative tumors may be reflec- available immunoassay assay against well-characterized tive of biological processes that differ from HPV-positive epitope(s) of LAMC2 that can be used to test the presence tumors. HPV-negative OSCC is more frequently associ- of LAMC2 in saliva is not yet available. ated with smoking, and tobacco smoke has been shown to induce MMP1 mRNA expression in fibroblasts (46, 47) Conclusion and skin (48). However, we did not observe increased MMP expression associated with smoking for the cell Results from our investigations into MMPs lend sup- types in our data (results not shown), and the reasons for port to the importance of this family of genes in the higher MMP expression in HPV-negative tumors remain pathogenesis and progression of OSCC. The very strong unknown. Despite these results, the difference in association of MMP1 with OSCC in particular provides a observed means for gene expression and salivary protein robust candidate marker for further investigations into its concentrations compared with controls was still substan- use as a potential marker for OSCC development. The tial within both site and HPV status categories, supporting strong associations of MMP salivary protein concentra- the use of these MMPs as biomarkers, regardless of site or tions with OSCC warrant further investigations into their HPV infection status. use as salivary biomarkers. We previously observed LAMC2 expression level to be ¼ highly predictive of OSCC-specific survival (AUC 80%, Disclosure of Potential Conflicts of Interest CI: 69–91 for LAMC2 combined with stage; ref. 49); elevated expression of MMP1 and MMP3 observed in this No potential conflicts of interest were disclosed. study was not as strongly associated with poor survival as LAMC2. However, we did observe a moderate elevation in Acknowledgments risk of all-cause and OSCC-specific death associated with We thank the study participants and their families and acknowledge the expression levels of MMP1 and MMP3. However, when resources from and use of facilities at the VA Puget Sound Health Care MMP1 and MMP3 were mutually adjusted, there was not System, University of Washington Medical Center and Harborview Med- as much evidence of their association with prognosis. ical Center, Seattle, WA. Using more stringent filtering criteria and a different normalization algorithm, we previously identified 131 Grant Support probe sets (111 unique genes) with mRNA expression U.S. NIH (R01CA095419 from the National Cancer Institute), NIH that were substantially different between OSCC cases and 5T32DE007132-28 (Comprehensive training in inter-disciplinary oral controls (25). The top predictive model comprised of the health research), and institutional funds from the Fred Hutchinson Cancer Research Center. genes LAMC2 and COL4A1 was able to distinguish cases The costs of publication of this article were defrayed in part by the from controls almost perfectly in our testing set (AUC ¼ payment of page charges. This article must therefore be hereby marked 99.8%) and an independent data set (AUC ¼ 97.6%). advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Taken together, these results suggest that markers identified using different statistical analyses may have the Received May 27, 2011; revised September 8, 2011; accepted September same or similar ability to predict phenotype and represent 23, 2011; published OnlineFirst September 28, 2011.

References 1. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates 8. Nagata M, Fujita H, Ida H, Hoshina H, Inoue T, Seki Y, et al. Identi- of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer fication of potential biomarkers of lymph node metastasis in oral 2010;127:2893–917. squamous cell carcinoma by cDNA microarray analysis. Int J Cancer 2. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J 2003;106:683–9. Clin 2010;60:277–300. 9. Kurahara S, Shinohara M, Ikebe T, Nakamura S, Beppu M, Hiraki A, 3. Poh CF, Zhang L, Anderson DW, Durham JS, Williams PM, Priddy RW, et al. Expression of MMPS, MT-MMP, and TIMPs in squamous cell et al. Fluorescence visualization detection of field alterations in tumor carcinoma of the oral cavity: correlations with tumor invasion and margins of oral cancer patients. Clin Cancer Res 2006;12:6716–22. metastasis. Head Neck 1999;21:627–38. 4. Groome PA, Schulze K, Boysen M, Hall SF, Mackillop WJ. A compar- 10. Rutter JL, Mitchell TI, Buttice G, Meyers J, Gusella JF, Ozelius LJ, et al. ison of published head and neck stage groupings in carcinomas of the A single nucleotide polymorphism in the matrix metalloproteinase-1 oral cavity. Head Neck 2001;23:613–24. promoter creates an Ets binding site and augments transcription. 5. Lee YH, Wong DT. Saliva: an emerging biofluid for early detection of Cancer Res 1998;58:5321–5. diseases. Am J Dent 2009;22:241–8. 11. Ye S, Eriksson P, Hamsten A, Kurkinen M, Humphries SE, Henney AM. 6. Charous SJ, Stricklin GP, Nanney LB, Netterville JL, Burkey BB. Progression of coronary atherosclerosis is associated with a common Expression of matrix metalloproteinases and tissue inhibitor of metal- genetic variant of the human stromelysin-1 promoter which results in loproteinases in head and neck squamous cell carcinoma. Ann Otol reduced gene expression. J Biol Chem 1996;271:13055–60. Rhinol Laryngol 1997;106:271–8. 12. Zinzindohoue F, Blons H, Hans S, Loriot MA, Houllier AM, Brasnu D, 7. Ziober AF, Patel KR, Alawi F, Gimotty P, Weber RS, Feldman MM, et al. et al. Single nucleotide polymorphisms in MMP1 and MMP3 gene Identification of a gene signature for rapid screening of oral squamous promoters as risk factor in head and neck squamous cell carcinoma. cell carcinoma. Clin Cancer Res 2006;12:5960–71. Anticancer Res 2004;24:2021–6.

www.aacrjournals.org Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 2635

Stott-Miller et al.

13. Vairaktaris E, Yapijakis C, Derka S, Serefoglou Z, Vassiliou S, Nkenke 31. Visse R, Nagase H. Matrix metalloproteinases and tissue inhibitors of E, et al. Association of matrix metalloproteinase-1 (-1607 1G/2G) metalloproteinases: structure, function, and biochemistry. Circ Res polymorphism with increased risk for oral squamous cell carcinoma. 2003;92:827–39. Anticancer Res 2007;27:459–64. 32. Kleiner DE, Stetler-Stevenson WG. Matrix metalloproteinases and 14. A single nucleotide polymorphism in the matrix metalloproteinase-1 metastasis. Cancer Chemother Pharmacol 1999;43:S42–51. promoter enhances oral squamous cell carcinoma susceptibility in a 33. Rosenthal EL, Matrisian LM. Matrix metalloproteases in head and neck Chinese population. Oral Oncol 2006;42:32–8. cancer. Head Neck 2006;28:639–48. 15. Kainuma K, Katsuno S, Hashimoto S, Oguchi T, Suzuki N, Asamura K, 34. Egeblad M, Werb Z. New functions for the matrix metalloproteinases in et al. Differences in the expression of genes between normal tissue and cancer progression. Nat Rev Cancer 2002;2:161–74. squamous cell carcinomas of head and neck using cancer-related 35. Kerkela€ E, Saarialho-Kere U. Matrix metalloproteinases in tumor pro- gene cDNA microarray. Acta Otolaryngol 2006;126:967–74. gression: focus on basal and squamous cell skin cancer. Exp Dermatol 16. Suhr ML, Dysvik B, Bruland O, Warnakulasuriya S, Amaratunga AN, 2003;12:109–25. Jonassen I, et al. Gene expression profile of oral squamous cell 36. Spurr NK, Gough AC, Gosden J, Rout D, Porteous DJ, van Heyningen carcinomas from Sri Lankan betel quid users. Oncol Rep 2007;18: V, et al. Restriction fragment length polymorphism analysis and 1061–75. assignment of the metalloproteinases stromelysin and collagenase 17. Yen CY, Chen CH, Chang CH, Tseng HF, Liu SY, Chuang LY, et al. to the long arm of chromosome 11. Genomics 1988;2:119–27. Matrix metalloproteinases (MMP) 1 and MMP10 but not MMP12 are 37. White LA, Maute C, Brinckerhoff CE. ETS sites in the promoters of the potential oral cancer markers. Biomarkers 2009;14:244–9. matrix metalloproteinases collagenase (MMP-1) and stromelysin 18. Ye H, Yu T, Temam S, Ziober BL, Wang J, Schwartz JL, et al. (MMP-3) are auxiliary elements that regulate basal and phorbol- Transcriptomic dissection of tongue squamous cell carcinoma. BMC induced transcription. Connect Tissue Res 1997;36:321–35. Genomics 2008;9:69. 38. Dorr€ S, Lechtenbohmer€ N, Rau R, Herborn G, Wagner U, Muller-€ 19. Baker EA, Leaper DJ, Hayter JP, Dickenson AJ. The matrix metallo- Myhsok B, et al. Association of a specific haplotype across the genes proteinase system in oral squamous cell carcinoma. Br J Oral Max- MMP1 and MMP3 with radiographic joint destruction in rheumatoid illofac Surg 2006;44:482–6. arthritis. Arthritis Res Ther 2004;6:R199–207. 20. Lallemant B, Evrard A, Combescure C, Chapuis H, Chambon G, Raynal 39. Hui A, Min WX, Tang J, Cruz TF. Inhibition of activator protein 1 activity C, et al. Clinical relevance of nine transcriptional molecular markers for by paclitaxel suppresses interleukin-1-induced collagenase and stro- the diagnosis of head and neck squamous cell carcinoma in tissue and melysin expression by bovine chondrocytes. Arthritis Rheum 1998;41: saliva rinse. BMC Cancer 2009;9:370. 869–76. 21. Jordan RC, Macabeo-Ong M, Shiboski CH, Dekker N, Ginzinger DG, 40. Hu S, Arellano M, Boontheung P, Wang J, Zhou H, Jiang J, et al. Wong DT, et al. Overexpression of matrix metalloproteinase-1 and -9 Salivary proteomics for oral cancer biomarker discovery. Clin Cancer mRNA is associated with progression of oral dysplasia to cancer. Clin Res 2008;14:6246–52. Cancer Res 2004;10:6460–5. 41. St John MA, Li Y, Zhou X, Denny P, Ho CM, Montemagno C, et al. 22. Hack CJ. Integrated transcriptome and proteome data: the challenges Interleukin 6 and interleukin 8 as potential biomarkers for oral cavity ahead. Brief Funct Genom Proteom 2004;3:212–9. and oropharyngeal squamous cell carcinoma. Arch Otolaryngol Head 23. Brockmann R, Beyer A, Heinisch JJ, Wilhelm T. Posttranscriptional Neck Surg 2004;130:929–35. expression regulation: what determines translation rates? PLoS Com- 42. Shpitzer T, Bahar G, Feinmesser R, Nagler RM. A comprehensive put Biol 2007;3:e57. salivary analysis for oral cancer diagnosis. J Cancer Res Clin Oncol 24. Nagler R, Bahar G, Shpitzer T, Feinmesser R. Concomitant analysis of 2007;133:613–7. salivary tumor markers–a new diagnostic tool for oral cancer. Clin 43. Shpitzer T, Hamzany Y, Bahar G, Feinmesser R, Savulescu D, Borovoi Cancer Res 2006;12:3979–84. I, et al. Salivary analysis of oral cancer biomarkers. Br J Cancer 25. Chen C, Mendez E, Houck J, Fan W, Lohavanichbutr P, Doody D, et al. 2009;101:1194–8. Gene expression profiling identifies genes predictive of oral squamous 44. Mork J, Lie AK, Glattre E, Hallmans G, Jellum E, Koskela P, et al. Human cell carcinoma. Cancer Epidemiol Biomarkers Prev 2008;17:2152–62. papillomavirus infection as a risk factor for squamous-cell carcinoma 26. Lohavanichbutr P, Houck J, Fan W, Yueh B, Mendez E, Futran N, et al. of the head and neck. N Engl J Med 2001;344:1125–9. Genomewide gene expression profiles of HPV-positive and HPV- 45. D'Souza G, Kreimer AR, Viscidi R, Pawlita M, Fakhry C, Koch WM, et al. negative oropharyngeal cancer: potential implications for treatment Case-control study of human papillomavirus and oropharyngeal can- choices. Arch Otolaryngol Head Neck Surg 2009;135:180–8. cer. N Engl J Med 2007;356:1944–56. 27. Pyeon D, Newton MA, Lambert PF, den Boon JA, Sengupta S, Marsit 46. Kim H, Liu X, Kohyama T, Kobayashi T, Conner H, Abe S, et al. CJ, et al. Fundamental differences in cell cycle deregulation in human Cigarette smoke stimulates MMP-1 production by human lung fibro- papillomavirus-positive and human papillomavirus-negative head/ blasts through the ERK1/2 pathway. COPD 2004;1:13–23. neck and cervical cancers. Cancer Res 2007;67:4605–19. 47. Yin L, Morita A, Tsuji T. Alterations of extracellular matrix induced by 28. GeneChip Expression Analysis; Data Analysis Fundamentals. Affyme- tobacco smoke extract. Arch Dermatol Res 2000;292:188–94. trix. http://www.affymetrix.com/support/downloads/manuals/data_ 48. Lahmann C, Bergemann J, Harrison G, Young AR. Matrix analysis_fundamentals_manual.pdf. Accessed October 11, 2011. metalloproteinase-1 and skin ageing in smokers. Lancet 2001; 29. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, 357:935–6. et al. Bioconductor: open software development for computational 49. Mendez E, Houck JR, Doody DR, Fan W, Lohavanichbutr P, Rue TC, biology and bioinformatics. Genome Biol 2004;5:R80. et al. A genetic expression profile associated with oral cancer identifies 30. Partek Genomics Suite. Partek Inc. http://www.partek.com/partekgs. a group of patients at high risk of poor survival. Clin Cancer Res Accessed October 11, 2011. 2009;15:1353–61.

2636 Cancer Epidemiol Biomarkers Prev; 20(12) December 2011 Cancer Epidemiology, Biomarkers & Prevention

Tumor and Salivary Matrix Metalloproteinase Levels Are Strong Diagnostic Markers of Oral Squamous Cell Carcinoma

Marni Stott-Miller, John R. Houck, Pawadee Lohavanichbutr, et al.

Cancer Epidemiol Biomarkers Prev 2011;20:2628-2636. Published OnlineFirst September 29, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/1055-9965.EPI-11-0503

Supplementary Access the most recent supplemental material at: Material http://cebp.aacrjournals.org/content/suppl/2011/09/27/1055-9965.EPI-11-0503.DC1

Cited articles This article cites 47 articles, 12 of which you can access for free at: http://cebp.aacrjournals.org/content/20/12/2628.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://cebp.aacrjournals.org/content/20/12/2628.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department Subscriptions at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cebp.aacrjournals.org/content/20/12/2628. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.